According to Samir Brahmachari, director of India's Institute of Genomics and Integrative Biology, the only factor holding India back from becoming a genome analysis superpower is computing capacity, so his institute is doing something about it. Last week, IGIB announced that it has selected Hewlett-Packard to install a 4-teraflop cluster at its facility in New Delhi.
When deployed at the end of the month, the system will be India's largest supercomputer, and could place the country in the Top500 list for the first time. But Brahmachari said that the benefits that the system will bring to India's nascent genomics community far outweigh number-crunching bragging rights.
"We believe that there's so much to do in genome analysis, that there are so many genomes that are still unannotated, but it's a huge amount of computational work," he said. "India can put thousands of people [to work on that], and that's the dream we have for [IGIB], that we'll have these sorts of activities."
India "has no shortage of IT people, so we can play the numbers game," he said. "All we need is the hardware." As part of the purchase, HP will also fund several fellowships to help IGIB attract "highly talented IT professionals" that IGIB could not ordinarily afford to pay through its government fellowship, Brahmachari said.
HP will deploy a 288-node HP Cluster Platform 3000 based on ProLiant servers with Intel Xeon processors, running Linux and XC System software. IGIB also purchased a 24-node ProLiant cluster running Linux and an Integrity Superdome server, along with 12 terabytes of StorageWorks Enterprise Virtual Array 5000 storage and an HP StorageWorks MSL6060 tape library.
Brahmachari said that IGIB plans to take "a mixed-bag approach" to its IT architecture, so it opted for both the cluster and the Superdome system. "We want to see how the [Superdome's] SMP [architecture] does," he said.
Financial terms of the agreement were not provided, but Indian news services reported last week that the Indian government has provided IGIB with a budget of 430 million rupees ($10 million) for its IT architecture, for which it spent about a third on the HP system.
Indian news services also reported that IGIB is considering a Blue Gene system from IBM as its next purchase. Brahmachari confirmed that while "the evaluation of Blue Gene has certain limitations because most biological systems are still not 512-processor problems," the institute is "definitely interested" in Blue Gene.
"We have negotiated, we are more or less there, [and] we are formally going to request IBM to process an export license, and once they are able to process the export license then it's very likely that we will be able to procure" the system, he said.
IBM officials could not be reached for comment before BioInform went to press.
Brahmachari stressed, however, that whatever other IT vendors IGIB may work with, "with HP we have a partnership."
In addition to providing the computational system, HP researchers will work with IGIB staff to optimize a number of their internally developed software applications for broader use. HP will also help IGIB attract "highly talented IT professionals who are otherwise paid very heavily," according to Brahmachari.
Financial details of the HP fellowships were not disclosed.
Dave Crowley, life and material sciences global business manager at HP, agreed that the deal with IGIB "is more a partnership than a traditional purchase and sale agreement." The fellowships, he said, "seemed like the best way to help IGIB advance their science goals, and get the most out of the technology investment that they made with HP."
Brahmachari said that IGIB currently has a headcount of around 60 people, but that the institute plans to grow to around 200 people by the end of the year. "We are looking globally to find faculty," he said.
While IT talent isn't hard to come by in India, Brahmachari said that it is still a challenge to find employees well-versed in both computational science and biology. In an effort to build its base of "hybrid people who are pretty comfortable in either of the two," IGIB has instituted a software-development method dubbed "two chairs, one screen" — a form of extreme programming in which a biologist and a computer scientist must share a single computer.
"First, you save the number of machines that you need," he said. "Plus, what happens is that they interact and eventually write what makes sense biologically."
Brahmachari likened India's growing bioinformatics base to a pond in which lilies are growing: "In the pond, a lily doubles every day. One lily becomes two, two become four. And in 30 days, the pond is full of lilies. When do you notice that the lilies are doubling? You don't notice on the 15th day, but on the 25th day, suddenly you notice that a part of the pond is full of lilies." IGIB, he said "has been doubling the lilies for the last six years, so we are creating the manpower, developing the experience in the country."
A number of other factors are driving India toward rapid growth in bioinformatics, Brahmachari said. One is the government's "commitment to contribute to the post-genome sequencing era."
As evidence of this priority, the country's Ministry of Science and Technology drafted a national biotechnology strategy in March that included several "strategic actions" in the area of bioinformatics [BioInform 4-11-2005].
The Ministry listed the creation of high-performance computing facilities among a number of factors that would help accelerate further bioinformatics development in the country, and set a goal to produce "50-100 quality PhDs, 500 MSc and 500 advanced diploma holders in bioinformatics every year," according to a Ministry statement.
The government did not provide details on its biotech funding plans, but the Ministry expects the strategy to lead to annual revenues of $5 billion and the creation of 1 million jobs by 2010.
In addition, Brahmachari pointed out, India's access to supercomputing technology was limited during the height of the Human Genome Project, due to US export restrictions on high-performance computing systems that were lifted in 2002. "In 2001, it was not possible to get a 4-teraflop machine because of the sanctions," he said. "Now we can get large machines, so things have changed."
HP's Crowley said that India's growing appetite for high-performance life science computing "certainly represents a business opportunity for HP that we hope to be able to take advantage of, and obviously we have with IGIB from a business perspective. But I think in terms of looking at India as a country that's also investing in its own future … their ability to acquire and implement and exploit big technology is good all around."
— Bernadette Toner ([email protected])